More features: formatting, actual chat, and show PDF

bhaskatripathi / pdfGPT

PDF GPT allows you to chat with the contents of your PDF file by using GPT capabilities. The most effective open source solution to turn your pdf files in a chatbot!

https://huggingface.co/spaces/bhaskartripathi/pdfGPT_Turbo

MIT License

6.85k stars 836 forks source link

More features: formatting, actual chat, and show PDF #16

Open dezerving opened 1 year ago

dezerving commented 1 year ago

I wanted to see if we could improve the current feature set. The following is what I have in mind:

Add formatting to answer (new line between each citation, make the citations go from [4] to [Page 4] and make them a different color so they stand out (making it easy to reference)
Have an actual chat-like interface (think humata, chatpdf)
Show the actual pdf side-by-side (and maybe the citations in the answers can link to the page in the pdf viewer making it easy to reference)

These features would be dope. I'm looking into working on it myself but I know there's people more talented that could look into this

bhaskatripathi commented 1 year ago

Thank you for your interesting suggestions. I have added all of them in my checklist and will soon comeup with the next release. I am also planning to integrate results from different sources like Alpaca, lLama and hopefully Google bard (I am in the wait list). Will come up with an improved interface soon.

dezerving commented 1 year ago

Thank you for your interesting suggestions. I have added all of them in my checklist and will soon comeup with the next release. I am also planning to integrate results from different sources like Alpaca, lLama and hopefully Google bard (I am in the wait list). Will come up with an improved interface soon.

Awesome. A couple other things that could be useful:

multi pdf support (and being able to chat with multiple pdfs)
implement redis as vector db (keeps things selfhosted, rather than having to use pinecone, chroma, etc)

Just as a question, is there currently a limit to the pdf (page # or actual size) that can be uploaded?

bhaskatripathi commented 1 year ago

no there is no limit. However, until the current code is optimized with a vectorDB there will be a tradeoff with the performance

bhaskatripathi commented 1 year ago

@dezerving - Made the first change for Page No [XX]

dezerving commented 1 year ago

@dezerving - Made the first change for Page No [XX]

Awesome! This + adding new lines between each citation would make readability of the answers much better (especially longer answers).

marc76900 commented 1 year ago

Have any of you coded anything on this? @dezerving @bhaskatripathi @deepankarm @Jeffrey95 @taherfattahi

magedhelmy1 commented 1 year ago

Seems like the project is dead.

bhaskatripathi commented 1 year ago

@magedhelmy1 - it is not dead, but I am not getting time these days. Will resume it from next month. I welcome you as a contributor.

Yiiipu commented 1 year ago

I build a simple Mac pdf reader based on this project and it has chat memories. It also shows pdf side by side. You can check it out: https://github.com/Yiiipu/SimplePDF

dezerving commented 1 year ago

I build a simple Mac pdf reader based on this project and it has chat memories. It also shows pdf side by side. You can check it out: Yiiipu/SimplePDF

This is really cool. I like the way that this looks. Any way to have it be a webapp so it can just be hosted as a container somewhere? Something like https://github.com/Anil-matcha/ChatPDF.